About 20 Paying Users, Not One Second Purchase

TL;DR

Since the first real payment on 21 June 2026, EventDrop has about 20 paying people, just under 400 euros gross, zero refunds and zero disputes. The cut-off date is 6 September 2026. The number that hurts is not in the revenue line: every one of those people paid exactly once, nobody paid a second time. This is not a growth story. It is a measurement story about a small product where the interesting values sit at the end of a chain and both of them are zero. I am writing down what I measured, what I am not allowed to conclude from it, and which three decisions came out of it anyway.

What EventDrop Is and Why I Count at All

EventDrop is a photo collection point for parties. The host creates an event, the guests scan a QR code, upload their photos, and everything lands in one shared gallery instead of seventeen phone cameras. It is one of the things I build and operate myself alongside the consulting work, visible in my product list.

I read the numbers live from Stripe and from the production database, not from a dashboard screenshot. The reason is unromantic: at these volumes every metric swings. A dashboard with a default seven-day window tells me a different story every day when the numbers are this small. So I query the state, write the date next to it and hold on to that. Every number here shares the same cut-off date, 6 September 2026, and the observation window starts on 21 June 2026, the day of the first payment where money actually moved. The sessions from February and March were hundred-percent vouchers; in Stripe they show an amount of zero. They do not belong in a revenue view, even if they would make the list look friendlier.

The Numbers, Without the Gloss

About 20 people, just under 400 euros gross, an average ticket just under 20 euros. Almost all payments fell into August alone, and almost all of the revenue with them. The strongest revenue week was in August, the volume peak one week later with five payments. Since 31 August it has been quiet, not a single payment since.

The split across the pricing tiers: the large majority Spotlight, a handful of Premiere, one single Flash for 5 euros. Studio, the top tier at 99 euros, has existed since 30 August and has never been sold.

What stands out is not the size. For a side product, a low three-digit amount is neither embarrassing nor remarkable. What stands out is the shape: a single month carries almost everything, nothing follows it, and nobody buys twice. For a product sold per event, a second purchase would be the first real signal. You have one wedding, a company party every year, a birthday annually. Zero repeat purchases across the whole period means either the window is still too short, or the product did not leave enough behind the first time for anyone to think of it again. I lean towards the second explanation, and the next section is why.

What Usage Means When Only Half of Them Start

There are a good two thousand uploads in the database in total. The median per paid event is a low double-digit number. Across the events that actually got going, it is three times that. The gap between those two values is this entire article.

At the end of August I audited that systematically and limited myself to a 60-day window so that future event dates could not distort the picture. The result: half of the paid events never had a single photo. The audit text is unfriendly and clear at that point: "Real data: 50% of paid events (9/18, 60 d) never got a photo, activation is the biggest gap."

The rest of the chain does not improve it. Half of the paid events arrived at all, a third got past ten photos, one single share into a social network, not one recap video, not one extension. The two zeros at the end of the chain explain what the revenue figure cannot: everything I built behind the actual core was never reached, because people stop before they get there.

Where guests do upload, by the way, it works well: the large majority stay anonymous and upload eight photos on average. So the mechanism is fine. What is missing is the path to it.

There is one temptation I have to fend off explicitly. Taken across all paid events, including those outside the window, the share without a photo looks friendlier, and a prettier number could be built out of that. I am not calculating it, because for several of those events the date simply had not happened yet. Only the 60-day evaluation holds up. A better number standing on a worse method is not a better number.

About twenty people paid and not one of them paid twice. That is not a revenue problem, that is an activation problem with an invoice next to it.

The Pricing Page Asked Me a Question I Could Not Answer

Spotlight cost 15 euros, Premiere 39. When I put the plan definitions side by side, both tiers carried the same number: twelve features, identical feature flags. The only difference between the tiers was quantities, meaning storage, guests, runtime.

That is the question the pricing page puts to every visitor and that I could not answer: why would anyone pay 24 euros more for exactly the same features? The customers answered unambiguously. Two out of three paid events were the cheap tier. The decision document puts it like this: "A customer who reads the pricing page has no reason for Premiere and buys accordingly: 14 of 21 paid events are Spotlight." That is not disloyalty, that is the correct decision given the available information.

The quantity limits that were supposed to justify the surcharge never touched anyone. The largest measured event needed 1.35 gigabytes of storage, which is 2.7 percent of the 50-gigabyte Premiere limit. The largest guest count was 57 against a limit of 200. The table for co-admins and helpers had exactly zero rows on 18 August; the feature was never used. One sentence from the document has stayed with me: "A guest limit that allows 200 when 57 guests were measured is neither a selling point nor a protection, it is only the risk of turning away a 250-guest event."

Out of that came Pricing v2: differentiation by capability instead of by quantity. And Flash left the shop. One single sale, 5 euros of revenue, and the event in question had zero uploads. A tier that costs a column in the comparison table, four lines in every translation file and a branch in every plan switch, and returns 5 euros, is negative on balance. Technically it stays, it just is not sold any more. Ending something because the numbers say so comes easier to me now than it used to.

The Channels I Imagined for Myself

August in analytics: a good 500 visitors, just under 90 percent of them on mobile, almost exclusively from Austria. And then the number I had to read twice. Almost all of the visitors came direct. No search traffic, practically no referrals.

That is a finding about me, not about the product. There never was a channel. There was a circle. Every single purchase came from reach I carried myself, out of conversations, messages, recommendations from people who know me. Everything I had been telling myself about acquisition was a story about twenty-odd people from my own circle.

The partner programme fits the same picture. I designed it in February, switched it live on 16 August, cold-contacted partners on 26 August, and on 30 August I ended the pilot and rebuilt the marketing surface. Commissions: zero. Attribution since 24 August: zero, every purchase in that period came direct. On launch day a gate also stood in the way that would have made every invitation code unredeemable, and the partner programme would have shipped silently dead with a green pipeline. I have written about that kind of false evidence at more length in Verification, Not Typing.

What I Am Taking Away

1. A pricing tier that only sells bigger numbers sells nothing. When two tiers carry the same twelve features, everybody buys the cheaper one. That is not a customer problem, it is a design problem on my side. The customer read the pricing page correctly.

2. Limits that never bind are decoration. 2.7 percent utilisation of a storage limit means that limit has never touched anyone. It protects nothing and sells nothing, it just sits in the table and costs maintenance. The same goes for a feature with zero rows in the database.

3. The first purchase is the smaller problem. When half of the paying customers never upload a photo, the funnel does not leak at the start, it leaks immediately after. I spent months thinking about acquisition while the loss happened three steps later. The next step is activation, not reach.

4. Almost everyone came direct, so there never was a channel. That share said more about my marketing assumptions than any conversion rate could. A product that sells exclusively through my own circle has not yet proven that it works outside of it.

5. A null result is a result. The partner programme produced zero commissions and still produced something solid: every check reporting green says nothing about whether anything reaches the user at all. That lesson cost more than the pilot did.

Conclusion

No pivot, no pathos. The numbers do not say the product is wrong. They say I was working in the wrong place. The people who paid never reached the core, and everything behind it, recap and extension, was therefore never even tested. The next round goes into activation: what happens between paying and the first uploaded photo, and why does half of it stall there.

Twenty-odd is too few for statistics. It is enough for a direction, once you stop rounding the numbers in your own favour. What the road from a weekend build to an operated product generally looks like for me is in From Prototypes to Product. What I take from this one is simpler: a measured twenty is worth more than an estimated two thousand.